Luotsi 0.1.0

What changed for agent builders

• Luotsi now has a stable first-minute output loop for agents: command -> structured output -> artifact root -> replay command -> next action.
• inspect and view --json keep JSONL session semantics explicit, while one-shot commands keep the luotsi-command.v1 envelope contract for scripts and CI.
• Failed runs can start with luotsi replay packet --artifacts <artifact-root> to write run-summary.json and run-summary.md, then use replay packet --check as the packet validation gate.
• The repo includes reusable agent guidance and parser examples for extracting the next command from envelopes, JSONL logs, and persisted run-summary packets.

What changed for engineers and CI

• The CLI is packaged as self-contained release archives for Windows, Linux, macOS x64, and macOS arm64, with installer scripts, SHA-256 checksums, and release asset attestations.
• First-run setup is centered on luotsi quickstart, luotsi doctor, and luotsi doctor --fix so adb readiness, device selection, live-view prerequisites, helper APK presence, and FFmpeg view extras are visible before deeper workflows.
• Scenario runs, shared-lab device claims, JUnit/JSON reports, replay artifacts, failure capsules, replay graphs, and artifact packages now share the same evidence-first handoff model.
• Live view sharing is intended for trusted lab or development networks only; --share-bind uses raw TCP today and does not provide TLS or authentication.

Open this first

• Docs hub: https://digablesolutions.github.io/luotsi/docs/
• First five minutes: https://digablesolutions.github.io/luotsi/docs/getting-started/first-five-minutes/
• AI agent workflows: https://digablesolutions.github.io/luotsi/docs/core-workflows/ai-agent-workflows/
• Installation: https://digablesolutions.github.io/luotsi/docs/getting-started/installation/
• Replay and artifacts: https://digablesolutions.github.io/luotsi/docs/core-workflows/replay-and-artifacts/

Output and replay handoff

• For source-tree validation, run luotsi help output to see the JSON envelope, JSONL session, artifact root, and replay mental model.
• When a release note mentions failed CI runs, agent handoffs, or artifact packets, point the first follow-up to luotsi replay packet --artifacts <artifact-root> and the validation gate to luotsi replay packet --artifacts <artifact-root> --check so evaluators get run-summary.json, run-summary.md, the At a Glance summary, primary failure, recommended next action, and the 60-second checklist. Use luotsi replay open --artifacts <artifact-root> --dry-run when a human needs the replay front door without launching a browser.

What's Changed

Other changes

• [codex] Add docs architecture diagrams by @slideep in #133
• Prepare first stable release gates by @slideep in #134

Full Changelog: v0.1.0-rc.12...v0.1.0